In pbs-assess/gfsynopsis: Data Synopsis Reports for PBS Groundfish

\clearpage

Plot Descriptions {#sec:plot-descriptions}

g <- readRDS(here::here("report/tech-report/ggplot-objects/petrale-sole.rds"))
sp <- "Petrale Sole"

checking_square <- geom_polygon(data = data.frame(x = c(400, 600, 600, 400),
  y = c(5500, 5500, 5700, 5700)), aes_string(x = "x", y = "y"),
  inherit.aes = FALSE, fill = "grey50", lwd = 1, col = "black")

This section provides captions for each of the visualizations that form the species-by-species pages. r sp is used as an example species for all plots except for commercial catch per unit effort maps where Pacific Cod is used.

\clearpage

Relative biomass index trends from surveys

# happens to have data for all the surveys:
dat_survey_index <- readRDS(here::here("report/data-cache-aug-2023/canary-rockfish.rds"))
dat_survey_index <- dat_survey_index$survey_index
survs <- c(
  "SYN WCHG", "SYN HS", "SYN QCS", "SYN WCVI",
  "HBLL OUT N", "HBLL OUT S",
  "HBLL INS N", "HBLL INS S",
  "IPHC FISS", "MSA HS", "MSSM WCVI")
survs <- tibble(surv_order = seq_along(survs), survey_abbrev = survs)
survey_descriptions <-
  semi_join(dat_survey_index, survs, by = "survey_abbrev") %>%
  select(survey_series_desc, survey_abbrev) %>%
  unique()
survey_descriptions <- bind_rows(survey_descriptions,
  tibble(survey_series_desc =
    "Hecate Strait Multispecies Assemblage Bottom Trawl",
     survey_abbrev = "MSA HS"))
x <- inner_join(survey_descriptions, survs, by = "survey_abbrev") %>%
  arrange(surv_order)
x$survey_series_desc <- gsub(" $", "", x$survey_series_desc)
surv_abbrev_text <- paste0(x$survey_abbrev, " = ", x$survey_series_desc, collapse = ", ")

(ref:survey-index-example) Example relative biomass index trends from trawl and longline surveys for r sp. Dots represent design-based mean estimates of relative biomass and vertical lines around the dots represent 95% bootstrap confidence intervals. Time series with a solid trend line and shaded ribbon for 95% confidence intervals represent an index that has been standardized with a spatiotemporal model (see Appendix A) \@ref(app:stitching). 'Mean CV' is the mean of the annual coefficients of variation (CVs), and 'Mean $+$ve sets' indicates the ratio of the mean number (across the years) of sets that captured the species of interest to the mean number of sets. All vertical axes are scaled between zero and the maximum upper confidence interval value for that survey. When both types of indices are shown, they are scaled to have the same geometric mean and the axes are set to encompass the model-based confidence intervals. For the MSSM WCVI, years before 2003 are shaded grey to indicate that catches are considered less reliable than modern data. For the HBLL OUT, HBLL INS, and IPHC FISS the values are abundance rather than biomass. r surv_abbrev_text. Also included are coastwide or survey-wide indices for SYN WCHG/HS/QCS/WCVI, HBLL OUT N/S, and HBLL INS N/S.

g$survey_index + ggtitle(en2fr("Survey relative biomass indices", french))

\clearpage

Maps of relative biomass from surveys

map_text <- "Note that the coast has been rotated 40$^{\\circ}$ to fit all the maps in the available space. Depth contours are shown at 100 m, 200 m, and 500 m."

(ref:survey-maps-cap) Example maps of relative biomass (or catch rate) from trawl and longline surveys from the latest available years of each survey for r sp. Shown are the synoptic trawl surveys (left), the outside hard bottom longline (HBLL OUT) surveys (middle), and the IPHC FISS (right). Individual sets are shown in the two left panels as faint crosses (if the species was not caught in that set), or circles with the area of the circle proportional to the species density from the set. Colour shading (except for the IPHC survey) indicates predictions from a spatial model that includes depth and depth squared as predictors as well as spatial random effects [Appendix E in @anderson2019synopsis]. The colour scale is fourth-root transformed to render a visual pattern similar to a log transformation without overemphasizing differences close to zero. The colour scale ranges from zero to the highest value within each map, and dark grey indicates regions where models could not be fit. The synoptic and HBLL maps show predicted biomass density throughout the survey domain. The IPHC map shows the raw unmodelled data for fixed station locations---stations without any observations for a given species are shown as empty circles. Years on the left side of each plot indicate the year of the respective survey. Surveys (except IPHC) in which less than 2\% of the sets contained the species are not modeled and are shown with raw data only. Mean values shown at the bottom are the mean fish density values from the raw data for the entire coast for the indicated years; for the IPHC data the units are fish per effective skate, where an effective skate represents 100 circle hooks with 18-foot spacing [Appendix G in @anderson2019synopsis]. r map_text

trans <- "fourth_root_power"
units_transformation <- (1000 * 1000) / 1000
breaks <- c(0.002, 0.05, 0.3)

labs1 <- if (french) {
  "Densité\nprévue de\nla biomasse\n(kg/km^2)"
} else {
  "Predicted\nbiomass\ndensity\n(kg/km^2)"
}

labs2 <- if (french) {
  "Densité\nprévue de\nla biomasse\n(poisson/km^2)"
} else {
  "Predicted\nbiomass\ndensity\n(fish/km^2)"
}

labs3 <- if (french) {
  "Taux de\nprise\n(poisson/raie\nréelle)"
} else {
  "Catch\nrate\n(fish/effective\nskate)"
}
p1 <- g$survey_spatial_syn +
  labs(fill = labs1, parse = TRUE) +
  ggplot2::scale_fill_viridis_c(trans = trans, option = "C", breaks = breaks,
    labels = breaks * units_transformation) +
  guides(fill = guide_colorbar(), size = "none") +
  ggtitle(en2fr("Synoptic survey biomass", french))

p2 <- g$survey_spatial_hbll +
  labs(fill = labs2, parse = TRUE) +
  ggplot2::scale_fill_viridis_c(trans = trans, option = "C", breaks = c(2, 10, 50)) +
  guides(fill = guide_colorbar(), size = "none") +
  ggtitle(en2fr("HBLL OUT survey biomass", french))

p3 <- g$survey_spatial_iphc +
  labs(colour = labs3, parse = TRUE) +
  guides(colour = guide_colorbar(), size = "none") +
  ggplot2::scale_fill_viridis_c(trans = "fourth_root_power", option = "C",
    na.value = 'white', breaks = c(0.2, 0.4, 0.7)) +
  ggplot2::scale_colour_viridis_c(trans = "fourth_root_power", option = "C",
    na.value = 'grey35', breaks = c(0.2, 0.4, 0.7)) +
  ggtitle(en2fr("IPHC survey catch rate", french))

gridExtra::grid.arrange(
  p1,
  p2,
  p3, # + checking_square,
  nrow = 1)

\clearpage

Commercial fishery catches

(ref:catches-cap) Example commercial fishery catch plots for r sp. Catch from various gear types is indicated by colour shading. Catch is calculated as the summed weight of landings aggregated by year. Discards include reported discard weights from all fisheries combined; however, bottom trawl discards are considered less reliable prior to 100% observer coverage in 1996 and trap, hook and line, midwater trawl, and Strait of Georgia bottom trawl discards are less reliable prior to fisheries integration in 2006 and are therefore not included. Years before 1996 and 2006 are shaded grey to indicate that catches are considered less reliable than modern data: an at-sea observer program was implemented for bottom and midwater trawl fleets in outside waters in 1996 and an at-sea observer program was implemented for non-trawl sectors in 2006. Management areas, as indicated in the top left corner of each panel, are shown in Figure \@ref(fig:management-map). The at-sea observer program stopped in early 2020 due to COVID-19 and an electronic monitoring system was later developed in its place.

g$catch + ggtitle(en2fr("Commercial catch", french)) + labs(y = paste(en2fr("Catch", french), " (t)"))

\clearpage

Commercial bottom trawl catch per unit effort indices

(ref:trawl-cpue-cap) Example commercial bottom trawl catch per unit effort (CPUE) trends, with effort as hours trawled, for r sp. Solid lines represent CPUE trends standardized with a Tweedie GLMM (generalized linear mixed effects model) for depth, latitude, DFO locality region, vessel, and month of year [Appendix D in @anderson2019synopsis]. The line itself represents the estimate and the shaded ribbon represents a 95\% confidence interval. The dashed line represents an unstandardized commercial CPUE index calculated as the sum of catch divided by the sum of effort each year. Standardized time series are scaled to have the same maximum 95\% confidence interval. Unstandardized time series are scaled to have the same geometric mean as the standardized time series. These are relative index values---the absolute value of the time series is not useful because it depends on arbitrary levels that the standardization variables are set to. Management areas, as indicated in the top left corner of each panel, are shown in Figure \@ref(fig:management-map).

g$cpue_index + ggtitle(en2fr("Commercial bottom trawl CPUE", french))

\clearpage

Maps of commercial catch per unit effort

(ref:cpue-maps-cap) Example commercial trawl and commercial hook-and-line catch-per-unit-effort maps for Pacific Cod (note this figure is not r sp). Lighter shading indicates higher levels of a geometric mean of catch per unit effort in a given hexagonal cell. The colour scale is fourth-root transformed to render a visual pattern similar to a log transformation without overemphasizing differences close to zero. Cells are 7 km wide and are only shown in cases where there are at least 3 unique vessels in a given cell to meet privacy requirements. For bottom trawl, catch per unit effort is calculated as the weight of catch (landings plus discards) divided by hours fished for all positive tows from the groundfish trawl sector. Trawl data are shown from 2013 onwards after the trawl footprint was frozen [@wallace2015]. Trawl data from 2007--2012 are indicated as outlined light grey hexagons to illustrate fishing prior to the frozen footprint. For hook and line, catch per unit effort is shown as the number of fish recorded as landed or discarded per set. Hook-and-line data are shown from 2008 onwards. Including as many years of data as possible reduces the number of discarded fishing events when implementing the 3-vessel privacy requirement. r map_text

trans <- "fourth_root_power"

g_alt <- list()
g_alt$cpue_spatial <- readRDS(here::here("report/tech-report/ggplot-objects/pacific-cod-cpue-spatial.rds"))
g_alt$cpue_spatial_ll <- readRDS(here::here("report/tech-report/ggplot-objects/pacific-cod-cpue-spatial-ll.rds"))

p1_trawl <-
  g_alt$cpue_spatial +
  (if(french){
    labs(fill = "Moyenne\ngéométrique des\nCPUE du chalut (kg/h)")
      }
    else{
      labs(fill = "Geometric\nmean of trawl\nCPUE (kg/h)")
      }) +
  (if(french){
    labs(colour = "Moyenne\ngéométrique des\nCPUE du chalut (kg/h)")
      }
  else{
    labs(colour = "Geometric\nmean of trawl\nCPUE (kg/h)")
    }) +
  xlab(en2fr("Easting", french)) +
  ylab(en2fr("Northing", french)) +
  ggplot2::scale_fill_viridis_c(trans = trans, option = "D", breaks = c(2, 10, 50, 200)) +
  ggplot2::scale_colour_viridis_c(trans = trans, option = "D", breaks = c(2, 10, 50, 200)) +
  guides(fill = ggplot2::guide_colorbar(), colour = ggplot2::guide_colorbar()) +
    gfplot::theme_pbs() + theme(legend.position = "bottom")


p2_ll <-
  g_alt$cpue_spatial_ll +
  (if(french){
    labs(fill = "Moyenne géométrique\nà CPUE H & L\n(poisson/ensemble)")
      }
    else{
      labs(fill = "Geometric\nmean of hook-and-line\n(fish/set)")
      }) +
  (if(french){
    labs(colour = "Moyenne géométrique\nà CPUE H & L\n(poisson/ensemble)")
      }
  else{
    labs(colour = "Geometric\nmean of hook-and-line\n(fish/set)")
    }) +
  xlab(en2fr("Easting", french)) +
  ylab(en2fr("Northing", french)) +
  ggplot2::scale_fill_viridis_c(trans = trans, option = "D", breaks = c(2, 5, 10, 20)) +
  ggplot2::scale_colour_viridis_c(trans = trans, option = "D", breaks = c(2, 5, 10, 20)) +
  guides(fill = ggplot2::guide_colorbar(), colour = ggplot2::guide_colorbar()) +
    gfplot::theme_pbs() + theme(legend.position = "bottom")

gridExtra::grid.arrange(
  p1_trawl, # + checking_square,
  p2_ll,
  nrow = 1)

\clearpage

Available biological samples {#sec:bio-samples}

gridExtra::grid.arrange(
  g$survey_samples + ggtitle(en2fr("Survey specimen counts", french)),
  g$comm_samples + ggtitle(en2fr("Commercial specimen counts", french)),
  nrow = 2
)

\clearpage

Length composition data

```r for survey abbreviations.")} g$lengths + ggtitle(en2fr("Length frequencies", french)) + labs(x = paste(en2fr("Length", french)," (cm)"), y = en2fr("Relative length frequency", french))

\clearpage

## Age composition data

(ref:ages-cap) Example age-frequency plot for `r sp`. Female fish are shown as
coloured (or black) circles and male fish are shown behind as light grey
circles. The total number of fish aged for a given survey and year are
indicated along the top of the panels. Diagonal lines are shown at five-year
intervals to facilitate tracing cohorts through time.
We plot the most recent 15 year window for which ageing data exist.
See Figure \@ref(fig:survey-index) for survey abbreviations. Ageing precision plots
comparing precision of readings by two individuals ageing the fish are provided
for all species for which age data exist in Appendix A in
@anderson2019synopsis.

```r
g$ages +
  ggtitle(en2fr("Age frequencies", french)) +
    labs(y = en2fr("Age (years)", french))

\clearpage

Length-age and length-weight model fits

```r \left( \log(l_\mathrm{inf} (1 - \exp(-k (A_i - t_0)))), \sigma \right)$ where $L_i$ and $A_i$ represent the length and age of fish $i$, $l_\mathrm{inf}$, $k$, and $t_0$ represent the von Bertalanffy growth parameters, and $\sigma$ represents the scale parameter. The length-weight curve is of the form $\log (W_i) \sim$ Student-t $(df = 3, \log(a) + b \log(L_i), \sigma),$ with $W_i$ and $L_i$ representing the weight and length for fish $i$ and $\sigma$ representing the observation error scale. The degrees of freedom of the Student-t distribution is set to 3 to be robust to outliers. The variables $a$ and $b$ represent the estimated length-weight parameters. Female model fits are indicated as solid black lines and male model fits are indicated as dashed grey lines. Text on the panels shows the parameter estimates and open grey circles represent individual fish that the models are fit to. These figures include all survey samples. See Appendix H in Anderson et al. (2019) for details on the models.")} cowplot::plot_grid(g$vb + ggtitle(en2fr("Growth", french)) + labs(x = en2fr("Age (years)", french), y = paste(en2fr("Length", french), " (cm)")), g$length_weight + ggtitle(en2fr("Length-weight relationship", french)) + labs(x =paste(en2fr("Length", french), " (cm)"), y = paste(en2fr("Weight", french), " (kg)"), linetype = en2fr('Sex', french), colour = en2fr('Sex', french)), align = "v")

\clearpage

## Maturity frequency by month

```r
g$maturity_month + ggtitle(en2fr("Maturity frequencies", french))

\clearpage

Maturity ogives

cowplot::plot_grid(g$mat_age +
    ggtitle(en2fr("Age at maturity", french)) +
    labs(x = en2fr("Age (years)", french),
      y = en2fr("Probability mature", french)),
  g$mat_length +
    ggtitle(en2fr("Length at maturity", french)) +
    labs(x = paste(en2fr("Length", french), " (cm)"),
      y = en2fr("Probability mature", french),
      linetype = en2fr('Sex', french),
      colour = en2fr('Sex', french)),
  align = "v")